基于变换域的遥感测绘图像条带噪声滤波方法  

Strip Noise Filtering Method of Remote Sensing Mapping Image Based on Transform

作　　者：季金亮 马庆勋[1] 李紫微 JI Jin-liang;MA Qing-xun;LI Zi-wei(College of Surveying and Mapping Science and Technology,Xi'an University of Science and Technology,Shaanxi Xi'an 710054,China)

机构地区：[1]西安科技大学测绘科学与技术学院,陕西西安710054

出　　处：《计算机仿真》2023年第3期206-209,301,共5页Computer Simulation

摘　　要：为有效消除遥感测绘图像条带噪声,并最大限度保留图像细节信息,提出一种基于变换域的遥感测绘图像条带噪声滤波方法。构建电荷耦合元件辐射响应函数,归一化灰度输出值实现辐射响应非均匀性校准,将图像条带现象作为两类加性噪声构成的图像退化,组建信号退化模型,掌握图像条带噪声关键来源与分布状况;运用矩形低通滤波器计算噪声形态变换过程,得到遥感测绘图像中的高频信号频域;把图像从空间域变换至频率域,对间隔一个像素宽度的条带实施傅里叶变换,探寻条带噪声周期性分布特征,在初始图像全部奇数行灰度值为0的情况下,实现条带噪声滤波处理。实验结果证明,所提方法能够有效剔除遥感测绘图像中的条带噪声,很好地保存初始图像关键信息。In this paper,a band noise filtering method of remote sensing mapping image based on transform domain is proposed for eliminating the band noise of remote sensing mapping image and improving the integrity of image detail information.The radiation response function of the charge-coupled element was established to de-normalize the gray output value in order to calibrate the radiation response non-uniformity.The signal degradation model was established,and the key sources and distribution of image stripe noise were investigated.The rectangular low-pass filter was applied to calculate the noise shape transformation process,obtaining the frequency domain of high-frequency signals in remote sensing mapping images.The image was converted from the spatial domain to the frequency domain.Based on Fourier transform,bands with an interval of one pixel width were transformed to explore the periodic distribution characteristics of band noise.When the gray value of all odd rows of the initial image was O,the stripe noise was filtered.The experimental results show that this method can not only effectively eliminate the strip noise in the remote sensing mapping image,but also save the key information of the initial image.

关 键 词：变换域 遥感测绘图像 条带噪声 噪声滤波 傅里叶变换 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]